EEVblog #1089 - XTAL Oscillator Teardown

hi it's time for a quick two-minute teardown we're gonna tear down a crystal oscillator no this is not a crystal oscillator this is a crystal I'm in the classic heat c49 you packages she'll be familiar with you see that it's just like a soldered shut at the end and it pulls out and there's photos of these everywhere in fact if you look on Wikipedia for crystal oscillator you'll see a photo that I actually added way back in 2006 inside one of these classic HC 49-year packages but this is not a crystal oscillator it's just a crystal it needs external circuitry to actually work so I thought we'd do a quick to me to tear down cuz I don't think I've ever seen inside one of these um this is a crystal oscillator in the standard 14 pin 4 pin dip package I know that sounds confusing but anyway those little bumps on the bottom by the way are to keep it off the PCB to stop the metal case here from actually shorting out and the pads or vias that you know exposed vias or anything else on the board just allow some standoff there anyway this is a complete oscillator it basically contains or will contain a crystal of some description they do come in many different types and physical form factors and things like that but this will also have in this case that's a hate CMOS oscillator circuit in there you've got there's a power P and a ground pin and a not connected or out like a tri-state enable pin and an output pin on it and it outputs a square wave it has a proper oscillator and an output buffer because this actually comes from the previous video where I actually had this CMOS oscillator on here and I did actually consider at the time well does this have a bypass capacitor built in because it turns out it worked just fine so let's have a look inside see what's in there is it a same a similar form factor as this with a little oscillator circuit in it I don't know let's find out Wow check it out if he's actually quite similar to the circular quartz discs that you get inside the US standard hey c49 you packages or a lot of them and then we've just got a standard 8 pin dip oscillator / driver whatever that is I don't know it's their own brand they've rolled their own or they've rebase it or whatever there's the quartz disc mounted in three places which is interesting Wow okay the ones in the side the HC 49-year package only have the two mounts on them this one has a third mount over here wonder why if that's not fr4 fiberglass that's actually a ceramic hybrid base on there although it's not really a hybrid because this is just a ceramic PCB hybrid would be like that actually embed the resistors on there and stuff like that but they haven't done that that's a zero ohm jumper here you go that's not even a resistor that's it no no there we go it note is one cap there's a cap there is a bypass cap is it there's one under there it's going to pin eight of the chip here which also goes along that trace up to the power pin here and it looks like the other side of it likely goes under the chip and connects to there I can just bust it out to confirm but I'm sure it does so this thing does have a bypass cap so there you go but that ultimately didn't interfere with my test for in the previous video for the bypass capacitors but there was one in there because this is the exact model that I used in there I had multiple ones of this as a bonus let's do one of these smaller eight pin dip ones mm-hmm and I guess I'd say all this circuitry here is going to be packed underneath the quartz disk there so it's just gonna be like stacked up yep I was right it was pretty obvious and oops yeah they're very close to the top of the I can on there so what I got my Dremel in there it just shattered cuz these things shatter really easily but it just suspends it over the top there but check this out it's totally different to the other one look at those mounting posts on there they're actually Springs on both sides there and there is no our third mounting point like we saw on the previous one so that's really interesting in fact you could probably come a guts in there if you weren't careful because if the and if the crystal will form some sort of resonant mode with a vibration mode with the spring and you could be in trouble but that's that's quite fascinating how they've actually added this Springs in there because quartz crystal oscillators and I've done quite a bit of research into this in a former job are very susceptible to shock and vibration in fact I've done some research actually shocking crystal oscillators I built a jig to actually drop them and shock them and get the response when put accelerators IO cellar ah meters on them and to measure a response and you would actually reset the drift characteristic of the oscillator when you actually shocked it so like Inlet like it's very small amount it's very marginal but if you're designing high stability oscillators which we were for underwater seismic stuff the stability is what mattered over time and if they got shocked then it would actually reset that drift characteristic and you'd have to start your drift compensation stuff all over again it just ruined everything so I just find that fascinating that they've got those Springs on there are they they're trying to eliminate sharp shocks which of course are directly coupled through the pins on the board straight into the ceramic PCB and then straight up the shaft onto the quartz are plates and they're trying to avoid that need so exactly the same as the HC 49 crystal again with a bypass cap between ground the ground pin and what looks like I'll buzz it out but I'm sure it is the power pin up there so these crystal oscillators I wouldn't take it as an absolute rule but two out of two have bypass caps in them but they don't mention this on the datasheet that I've seen anyway so what effect does this bypass capacitor have on here on our previous experiment that we did where I face bypass capacitors on the board and show the effectiveness of them I'll link that in at the end if you haven't seen it well it does have some effect it doesn't negate the previous video in any way it's just that the effect would have been more dramatic last time if we didn't have that bypass capacitor inside this thing but you've got to remember though that that bypass capacitor is inside this thing which has those little tracers running we've got the leads on here we've got the inductance of these lead links where they've got the inductance of the traces all going through there and of course our reference plane here that we're actually measuring everything relative from is outside the package so having the bypass capacitor inside the package it's not as it's not the same as having it actually directly connected to the ground plane the reference that we're actually measuring from which is what we actually care about in this case it's going to have an inductor it basically is almost practically no DC resistance there but it's going to have inductance of the leads and the traces and everything else on that path no matter how small they are they're still going to have a dramatic effect at the frequencies we're talking about it so that bypass capacitor has a nice effect on this chip of course it really helps a lot but outside the circuit well it's going to help too but relative to the reference plane it's just a different thing so anyway what I've done is I managed it I couldn't get in there off the solder and I hope I managed to get in there with a little flat headed screwdriver and just prise it out the capacitor just crack it because they're ceramic capacitors and they crack out really easily I was able to do that without damaging the quartz resonator there so it still works but we have no bypass capacitor on there so let's take another look now I won't go over the whole setup again you have to watch the previous video to get an idea for that so it's still got being on one megahertz here so we didn't physically damage the resonator there but look at the ripple that we're getting now rather than just at the edge there it's got all this other crap in here as well at multiple points inside that one the one megahertz fundamental so that's rather interesting is it and look at channel 2 we're now two volts per division on here this is crazy so it's like the five volt rail is just going up by two and a half down by two and a half that's just ridiculous that's with absolutely no bypass capacitance that rail is horrid but of course that is driving our 50 ohm load which we had there before and yes I am using the proper probing that we did last time so if we disconnect our 50 ohm load we should see that improved dramatically so there you go that's with no load and you can see there's now no large transitions it's still very bad on the five volt rail the blue channel here at two volts per division but no large transitions that we saw before and if I put one of the resistors back you'll see it drops in amplitude and you get that large transition going there like that so that's actually sync in high frequency current into the load with no bypass capacitor it doesn't have any bypass capacitors store that little gulp of energy that it needs so I really should have setup this video better last time but let's have a look what happens now if I whack a that 330 micro farad back on here still got no high-frequency bypass but if we do the bulk decoupling still does quite a reasonable job look at that but that high-frequency stuff at a hundred millivolts per division is still there and look at the large stuff on the positive transition here there we go from the positive transition you can really see it that's absolutely enormous so because it's got no high-frequency bypass capacitance but let's see if we can see this change as I slide it I'll start near there and I'll slide it backwards so here we go there we go look at that and I'll slight look at the level look at the level keep looking keep looking and you can see it going higher and higher level as I move that bypass capacitor so I've got it right down here and I put it up there and it makes quite a dramatic difference in terms of that's 100 millivolts per division in that high-frequency ripple even though we're not you really using a real optimized cap for that so let's try our point 1 microfarad ceramic shall we but at this point any capacitance is going to make a difference so even the big bulk cap up there's gonna do a reasonable job will find that this ceramic here is there we go that's near it and we move it away whoops move it away and it gets bigger and bigger yet again so let's use that 0.47 micro farad film cap on there there we go and now let's try and replace that with this point one ceramic probably won't see much difference cuz they're both gonna be there we go pretty much both equally effective there okay let's do the combo now point one and the naught point four seven come on there you go sweet us up stay there you mongrel it does have a bypass capacitor in there but that doesn't mean that you shouldn't use a bypass capacitor on that device because you've got the inductance of the leads and everything else so it's not as effective relative to the reference plane and when you're trying it when you've got a driver over here and you're trying to drive another chip over here and this chip all the chip cares about there is the what's actually received relative to this reference plane and the power plane here when you start adding little leads and everything unducted all in series and stuff like that you start to complicate the equation doesn't look great does it but just put your little bypass copy on there and she's sweet as look at that like a bought one now if you're wondering what all this stuff in here actually is obviously it's not just the one pulse and then just some ringing in there that then eventually settles out there's obviously some very deliberate high higher frequency components in here so if we set up some cursors here like roughly from one peak to the next that's sort of higher frequency stuff we're looking at about you know twenty nine point four megahertz or something like that but there's something more interesting which are these period higher Peaks in here like this so this isn't like just your normal ringing it wouldn't do that something is resonating or oscillating at that particular frequency it's giving it a kick each time so it's obviously oscillating something like that so if we move the cursor over there from one peak to the other uh-huh what are we got eight megahertz aha and you might have seen that in some previous footage here I might have to replay it and I think even the previous video that meant how well eventually we bypass them is pretty good but you could see this like a higher frequency like little spikes in there so let's actually go in and actually probe the crystal oscillator it's the actual crystal resonator inside there and see what frequency we get I think this in fact I'm pretty sure this is going to not use a 1 megahertz wrestle resonator it's going to use an 8 megahertz resonator and they're actually dividing that by eight and the chip in there might be have some pin straps to give you different frequencies different divider ratios for example so that might be how they might get the different frequencies out of the thing obviously it can't get like the oddball ones with the same resonate if you've got an 8 mega that's resonating you're not going to get you know the 2.40 for 8 megahertz for example probe 1 pin hopefully we were don't shut down the oscillator with the capacitance uh-huh what's the frequency down the bottom there eight megahertz there it is so there you go it's obviously an 8 megahertz resonator on here and divided by 8 and that's why you get that higher frequency stuff yeah there's a ringing in there but as you saw it gave an extra kick every time the 8 megahertz oscillator did its business so you might have to remember that when you're doing EMC compliance in the rest of it you're going to have to factor in in this particular case it's a divided by 8 so the actual see is 8 times higher and that sort of stuff can actually leak out if your pins into your ground planes and and actually radiate or a couple out so you know you've just got to be aware of that it's not a 1 megahertz oscillator it's actually an 8 megahertz / hey I know that was slightly more than two minutes but yeah I wanted to see inside these things I'd never actually cut one apart before and it's pretty much exactly as I expected so nothing hugely groundbreaking there but at least I know and now you know I hope you enjoyed that if you did please give it a big thumbs up as always subscribe at the end play the videos at the end here all that sort of jazz subscribe to eevblog too and as always there you can subscribe on patreon as well thanks to all my patreon subscribers who often by the way do get some but not all videos early before I release them on the main channel if you wonder how someone comments from a day ago before it was released well that's how they do it catch you next time and if you found that interesting - be sure to stick around for the links I've got two three videos at the end of this after the end screen here one is the crystal oscillator drift which I talked about in the circuitry to do that and how we did that back in the day that's a real old video and I've also got one on how to detect gravity using a frequency counter and it has to do with crystals it's a fascinating thing check that one out definitely and I've also done a rubidium frequency standard teardown as well check them all out

Channel: EEVblog

Views: 501,560

Keywords: eevblog, video, crystal oscillator, quartz crystal, what's inside, teardown, bypass capacitor, ceramic capacitor, decoupling capacitor, pcb ground plane, ground plane, emi, emc, emc conformity, emc radiation emission, emc testing, oscillator circuit, oscillator, xtal oscillator

Id: dkdLWronrlk

Length: 17min 2sec (1022 seconds)

Published: Mon May 28 2018